Exphygit Corpse “text” . . . The Wrong 2025-6

[ Noah Travis Phillips’

essay/text to accompany exphygit exquisite compost corpses 1-5

five solo “Exquisite Corpse” style collages for Exphygit Corpse

an exhibition in three stages, curated by Ricardo Bodini and Camila Jordan

for the “Compost^ Pavilion” as part of The Wrong Biennale of New Digital Art 2025-6. ]

The stages of the exhibition Exphygit Corpse :

Phase One: The Offering. Donation of Discarded Images: Through collaboration, we compile discarded AI-generated images from the participants.

Phase Two: Mycelium. These discarded images become the raw tissue for artistic exploration, where participants recycle and remix the fragments to craft a renewed aesthetic narrative, breathing new life into what was once deemed expendable.

Phase Three: Terra preta. A Space for Text Speculation. Possible contributions include, but are not limited to, the following: Micro-essays, critic reflection, Glitch haiku, Speculative fiction fragments, AI-related fears, dreams, confessions, Collaborative chain texts (exquisite corpse style) related to the Compost^ statement.^[1]

At the edge of a post-human threshold, we find ourselves in a moment that is more anthropocene^[2] than ever. In Exphygit Corpse^[3] we contemplate and encounter “the new artificial intelligence”, recognizing the apparent drastically increased implications and impact of large language models (LLMs), and a more urgent sense of the possibility of non-human and even non-organic learning (thinking, … training confidence, and hallucination.), and the fundamental invitation to think about (other) things “thinking”.^[4] The exhibition was sequenced in three phases (above). Artist Noah Travis Phillips participated in the full breadth of the exhibition, contributing their own compost or “B-sides”, using material from other artists in the exhibition to create five “exquisite corpses”, and writing this text.

In the first phase of the exhibition, The Offering, the artist gleaned and collected (downloaded) / appropriated materials from the “Compost^” collection^[5] – and in this same spirit, Phillips contributed hundreds of images from their own archive of synthesized images, created through research with an early GAN.^[6] Phillips refers to this kind of  experimental risk-taking, unconventional, and raw material as their “B-sides.”^[7]  Through this process of prompting,  generating, and collaging (visually & textually) Phillips as the artist must and can’t help but “internalize the knowledge contained in the image” as the images “make claims of their own” in a “complex interactive relationship” and are part of this “process of mutual influence … the object infects the person and something transfers from it to the person”. Particularly when working with GANs, “... appropriation becomes a process in which the artistic subject bargains with something that has unpredictable consequences … it generates dependency and amounts to a surrender to something. Being infected by something leads to a loss of control … artists who appropriate also subject themselves to the object.”^[8]  The algorithmic continues to produce and suggest, with some degree of authority, materials for paradigms and epistemological frameworks for understanding the world, political as they make the world appear in certain ways.

In the second Mycelium phase, Phillips created five "exquisite corpses” from this material, (re)composing and choreographing collaborations between material from others.^[9] 

Phillips treated the distinct and diverse styles or aesthetics they found as the multiple artists in a classic Surrealist “Exquisite Corpse”^[10], thus joining heterogeneity (of aesthetics, apps/platforms used, prompt style, country of origin, etc.), and working in a mode that highlights the modification of found materials, as part of an approach the artist often refers to as “made, found, & modified”. This strategy forwards the importance of editing and postproduction, and the use of not only these generative adversarial networks (GANs or text-to-image “AI” as instruments, but also the Studio itself as an instrument. As a GAN functions as a visual synthesizer, so too has the artist approached algorithmic systems and machine-learning (as tools and instruments) throughout a lifetime of creatively exploring the Internet. Engaging in dialogue with digital tools and algorithms has been central to their art practice since they were a teenager.

In the third phase of the exhibition, Terra preta, the artist produced this text (with the “help” of AI or other LLMs, it should be noted.) and in a way that again mimics “Exquisite Corpse” in its piecing together, and its rough connections between larger themes.

Phillips would suggest that the term “artificial intelligence” this is misleading^[11], and has previously called them “algorithmic systems”^[12], approaching and relating to them as tools or instruments. For Phillips, working with these systems as they do remains “[an] organic and intuitive response to increasing algorithmic (AI!) determinations”.^[13] For Phillips, the GANs producing these images function more as “visual synthesizers.” This technology has developed rapidly, and the GANs Phillips was most active with represent a moment that has passed. One specific GAN served as Phillips’ primary instrument^[14] (with one or two secondary or backups); now those GANs are defunct and can no longer generate new material. So, what remains is largely compost – leftovers, an archive, a folder on a desktop. “Exophytic” describes growth outward from the surface of origin, as this series extends from the artist’s own praxis.

~ ~ ~ ~ ~

 The GANs producing all this visual material, are more of a  “visual synthesizer”.^[15] This is and has been such a quickly developing cultural space. The GANs Phillips was initially and primarily interested in were a moment that has passed … and there was one specific GAN that was their primary instrument in all of this (with one or two secondaries and/or backups), and so this time has passed for me … and all I really have left is some kind of compost … or leftovers (an archive at best, a folder on my desktop).  Working with GANs and LLMs is a process of improvisation, a negotiation with systems that extend beyond human authorship – a distinctly posthuman perspective, and one that offers glimpses into our own patterns of thought, refracted and remixed through machine synthesis.^[16] 

We are eating ourselves, self-cannibalism, cultural cannibalx^[17] or what the artist might refer to as recycling & remix (again, “made, found, & modified”). After synthesis through algorithmic systems it is a dialogue w/ other-than-human paradigms (made and shaped by humans, initiated by humans?) Another environment we have turned into a situation of total information, recycling, recursion, and feedback, made by us and that we now exist within it (another possible definition of the Anthropocene). “Exquisite Corpse” is about juxtaposition … two unlikely things meet and in that meeting something unpredictable happens – this is like the meeting between people and “machine learning”.^[18] “AI”, GANs, LLMS, diffusion models, and other frameworks continue to reveal how the things we eat/consume/digest influence and shape us, and the things we do, and the art and culture that we make … These approaches are highly Postmodern in its recycling (of imagery, data, visual culture, &etc.) We feed ourselves back to ourselves … and so these “AI” become a source of supracultural feedback.  

This project continues an element within the artist’s praxis that explores imagination as an essential force that creates and transforms reality, rather than as a form of escape from the present.  With GANs we are “...immersed in and able to conjure visions / fever dreams of and from multiple realities, multiple possibilities, multiversal, …”^[19] as well as “unpredictable cognizances and understandings”^[20] … we are hallucinating with the machine, and within the machinic, we become (and have always-already been) hybrid, cyborg, chimera, mutant, monster, … (like Exquisite Corpses ourselves).

It requires a considerable amount of political, cultural, and historical context to fully understand art (and artifacts) made in this way …  contingent as it is in a multiplicity of personal, cultural, and technological variables and vectors. These “AI” instruments/tools/spaces and our current cultural moment of recognizing, relating to, navigating, them is rife/ripe with:

“... allegory, that literary and pictorial mode that, at one level, uses concrete images, characters, or landscapes to represent the abstract relationship between ideas …”

“Allegories are … a rather paradoxical way of explaining concepts with symbols. As the literary scholar Angus Fletcher points out, allegories often take place in fantastic, almost psychedelic environments - a dreamland, a visionary otherworld, or a futuristic scenario where magic appears as superscience. At the same time … allegories are usually dry and schematic, as they tend to follow abstract or logical relationships between concepts …”

“... some characters in allegorical texts are so programmatic that they were known… as “allegorical machines”.”

Erik Davis notes that “Angus Fletcher defines allegory as “a fundamental process of encoding our speech,” and computers are nothing if not hierarchies of encoded language.”^[21]

“Exquisite Corpses” are allegorical^[22] – collages creating a body/figure, a hybrid being/entity,  “...the more profound sense of being human-technology symbionts: thinking and reasoning systems whose minds and selves are spread across biological brain and nonbiological circuitry.”^[23]

The figures/characters have no backgrounds, so they stand on their own… function more as entities, and can enter different contexts (posthuman explorations). They barely hold together as entities. Their sense of singular, stable, autonomous subjectivity is highly questionable. Instead, they (as we) are socially, linguistically, temporally, and culturally constructed.

“The heteroglossia of data we are confronted with demands complex topologies of knowledge for a subject structured by multi-directional relationality. We consequently need to adopt non-linearity to develop cartographies of power that account for the paradoxes of the posthuman era.”^[24]

The posthuman reality revealed here is an example of  the ways that “...the distributed cognition of the emergent human subject correlates with — in [Gregory] Bateson’s phrase, becomes a metaphor for — the distributed cognitive system as a whole , in which “thinking” is done by both human and nonhuman actors.”^[25] And highlights how "The only way by which one could be sure that a machine thinks is to be the machine and to feel oneself thinking."^[26] which we begin to feel more poignantly the possibility for and edge of w/ current LLMs.  

~ ~ ~ ~ ~

Often experienced as a “… collation of natural elements and a physical body, a sense of trepidation and estrangement apparition, apprehension, reassemble and mutate …^[27] feeling almost “right” but not quite. This evokes an inhuman (posthuman) logic,

“... these unnerving visions suggest an uncanny realm where the familiar and unfamiliar are fused into an embryonic space of image production.” ^[28]

“...AI do image-processing models do not experience the world in phenomenological, embodied terms but replicate a once-removed and askew version of it … reveal how algorithms computationally generate disquieting allegories of our world.” ^[29]

an apparition (a nightmare?) where the physical, rational, ethical    is displaced (usurped?)  half-alien mecha-logic

inhuman processes – humanized (again as in a synth)

machinic ontologies of perception replacing the ocular^[30] How has the machinic already influenced and shaped how we see? already adumbrating and (over)determining what we experience and how …

“... [AI are] basically influencing machines, and they achieve influence through the extraction of value and the use of classification systems to define certain features of the world we live in. This is also where hallucinations come in: these influencing machines generate hallucinations, uncanny ways of seeing, that make you see something in a particular way or believe something in a particular way.”^[31] 

“AI” and Large Language Models continue to have confidence in things that are wrong / untrue^[32] and they hallucinate, hallucination is a form of glitch^[33] … Machines produce comprehensible results without comprehending anything themselves.^[34] They (“AI” and LLMs) can be thought of as averaging machines because its outputs are generated by finding the most probable, or "average," patterns and sequences in its training data.  We comprehend the results with a human intimacy, and it is our responsibility to amplify what is important to us in these prompt results and generations, to add our own “sauce and seasoning”. As always, we are always-already collaborating with other-than-human systems.

“Art has always existed in a complex, symbiotic and continually evolving relationship with the technological capabilities of a culture.  Those capabilities constrain the art that is produced, and inform the way art is perceived and understood by its audience.” ^[35] 

Lines and boundaries and borders between human thinking and computation begin to blur. Machines generate images humans are capable of understanding, and finding significance in. This means humans are not the only image-makers. Now we have also created synthesizers to do this for us, to help us to visualize.  

In our use of machine learning we are cannibalizing culture, in a way we are consuming ourselves. Previously, this kind of self-eating has led to madness. When we do something more with the material that comes from our interactions with algorithmic systems and machine learning tools, then we are working beyond the predictive modes and models perpetuated by machine-learning, which can only recycle (but so too do we). We are creating something fresh/strange; reiteration can be a powerful tool for creating something novel by rephrasing and restructuring ideas instead of simple repetition.

~ ~ ~ ~ ~

Phillips’ “Exquisite Corpses” (exphygit exquisite compost corpses 1-5) are … intentionally abstract, rather than reductive or deterministic – allusive and elusive,  as are hallucinations, the synthesized hallucinations they are collaged from. This is shaped in part by “humans have all sorts of taxonomies that we use to try to make sense of the world: taxonomies for dreams, tarot cards, historical events, ideas about things being “lucky” or “unlucky”, and even taxonomies of allegories. Humans’ relationships to images are fuzzy and malleable.”^[36] 

“the body is dismembered or reassembled, swollen or multiplied, propped with prosthetics or fused with nature and the machine … distorted and disoriented our most familiar of referents, playing out personal, cultural, or social anxieties and desires on unwitting anatomies.”^[37]

The artist has their own hallucinations with these instruments, as psychotechnologies^[38] – 

Exquisite Corpses are mutual(ly constituted) hallucinations. And in this way there is some of the spirit of the OG Surrealists here, in the way that things don't totally add up, and aren't necessarily answerable to one-another, this or that can turn into anything else, and be radically reimagined.^[39] The LLM results fray at the edges, where data and patterns joins human knowledge and wisdom. These softwares hallucinate; we can learn from the ways it reveals us to ourselves, as “Exquisite Corpse” and other Surrealist Games do, helping us to re-invent the world.^[40] As conceptual artist Sol Lewitt states in his “Paragraphs on Conceptual Art”: “The idea becomes the machine that makes the art.”^[41] In this sense, machine output doesn’t replace human imagination so much as extend and amplify – where intuition and pattern-seeking can become collaborative rather than predictive, a feedback instrument for cultural imagination showing how tools reflect and reshape our ways of seeing rather than dictating the future, and still might help us navigate a way forward. As in a cliche, to end with prognostication; unless we (as artists and free-thinking humans) do something to seriously instrumentalize these algorithmic systems and machine learning platforms and tools, they will continue to more deeply broadly and molecularly define our realities and our futures^[42] (which under current socio-political regimes continue to  enforce and normalize what sister bell hooks calls “imperialist white supremacist heteropatriarchy”^[43]. I have also explored a version of this hegemony, referring to it as techno-military-industrial-hypercapitalism”^[44]).

~ ~ ~ ~ ~

“Feminists have a long and rich genealogy in terms of pleading for increased visionary insight … with a strong critical and an equally strong creative function. That creative dimension has been central ever since and it constitutes the affirmative and innovative core of the radical epistemologies of feminism, gender, race and post-colonial studies. Faith in the creative powers of the imagination is an integral part of feminists’ appraisal of lived embodied experience and the bodily roots of subjectivity …”^[45]

It seems pertinent to approach these tools and instruments with a sense of valuing, respecting, and attending to our own human and creative perceptual and cognitive processes, and discovering when more humanity, more emotion, is wanted, beneficial, and necessary. “We also need to enlist affectivity, memory and the imagination to the crucial task of inventing new figurations and new ways of representing the complex subjects we have become.”^[46] By playing with these tools (when they/we are pragmatic) and instruments (when creative) like Surrealist Games we might cultivate a sense of unique self-knowledge and self-transformation from interaction with these predictive oracular) tools and practices^[47] as we do and have for centuries with other divinatory and creative technologies.

Artists feel (and are) called to do “something more” or “something else” with the world: with whatever materials, media, even images made through machine-learning tools, &etc. They work in dialogue with these tools and technologies. All tools have always contained biases – both in their design and in their use application. These frameworks aren’t always intentional, but they’re often unavoidable, because they’re created by humans with their own conscious and unconscious perspectives.^[48] If we’re living in an algorithmic bubble, we need to know how it’s bending and coloring whatever rays of light we’re able to glimpse through it.^[49] This kind of intervention feels crucial to any praxis-informed future with machine learning and related “AI”.